Everyone repeats "at least twice a week," but nobody tells you how many minutes that is. Before I looked into this, I had a picture in my head of an hour in the gym, two or three times a week.
It turns out somebody has measured the minutes. The story starts a lot lower down than I expected.
① The "twice a week" in the guidelines was not set by looking at death rates
Physical activity guidelines in several countries tell adults to do muscle-strengthening activity at least twice a week. But the evidence those recommendations rest on is mostly musculoskeletal health, meaning muscle and bone.
The authors of a 2022 meta-analysis said this outright. The frequency was not chosen by asking how much of it is linked to living longer.
So if you want an answer to "how much," the guidelines are not where it is. You have to go to the observational data.
② Doing some versus doing none shows up fairly consistently
That meta-analysis pooled 16 adult cohorts. The part comparing people who did no strength training with people who did any covered 7 studies, 42,133 deaths, and 263,058 participants.
In that comparison, the risk of dying from any cause was 15% lower.
The studies varied a lot among themselves. But the direction did not wobble: all 7 pointed the same way.
③ So how many minutes? A study published this year broke it into bands
A study out in June 2026 went straight at that question. It followed 147,374 people in three large US cohorts for up to 30 years.
What makes it different from earlier work is that it asked about strength training every two years and used the running average. That is closer to an actual habit than asking once and then following someone for three decades.
Below are the results after adjusting for aerobic activity, using people who did no strength training as the baseline. The figure in brackets is the average amount actually done by people in that band.
▸ 1–29 min/week (average 12 min) — 5% lower
▸ 30–59 min/week (average 42 min) — 9% lower
▸ 60–89 min/week (average 72 min) — 9% lower
▸ 90–119 min/week (average 102 min) — 13% lower
▸ 120+ min/week (average 204 min) — 8% lower
All five bands were statistically significant. In the lowest band, meaning people doing something like 12 minutes a week, a difference was already showing up.
That said, the estimated ranges for the bands overlap each other. So this data cannot tell you that one band beats another.
④ The two datasets split on the optimum, and agreed on the shape
The 2022 meta-analysis put the low point of the curve at 40 minutes a week. The 2026 cohort put it in the 90–119 minute band.
Two papers in the same journal, more than twice apart. Which is why this post cannot hand you a number for "the optimal amount."
What they do say together comes down to three things.
▸ A difference is already visible at very small amounts
▸ Doing more does not buy proportionally more
▸ At the top end the curve turns back up
Do not read that third point as "too much is harmful." Far fewer people sit in the high-volume bands, and the meta-analysis authors only said that region was unclear.
⑤ What these percentages are percentages of
"13% lower" does not mean your risk of dying dropped by 13 percentage points. It means 13% was taken off whatever risk you already had.
And that starting risk depends heavily on age. The median age in the 2026 study was 54, and most of the studies in the meta-analysis were middle-aged and older Americans.
The same study also looked at participants under 55 on their own. There were only 624 deaths there, and the estimates got shaky: the 1–29 minute band was significant while 30 minutes or more was not.
The grade of the evidence deserves saying out loud too. The 2022 meta-analysis rated its own all-cause mortality result as "very low" certainty, noted it could not test for publication bias because it had too few studies, and said its pooled estimates may therefore be inflated.
A 2025 umbrella review that gathered 48 meta-analyses points the same way. Not one association between physical activity and mortality reached either of the top two evidence grades.
⑥ The biggest movement was not strength training on its own
The 2026 study also crossed aerobic activity against strength training time. I think that table is the most useful thing in this post.
▸ People doing almost no aerobic activity — adding strength training barely moved the number
▸ People in the middle of the aerobic range — adding strength training produced the largest drop
▸ People doing a great deal of aerobic activity — the number was similar no matter how much strength training they did
Put simply: in this data, strength training did not substitute for walking or other aerobic activity. It made the biggest difference when it was added on top of it.
The umbrella review above also found its strongest association in people doing aerobic and strength work together.
⑦ So, depending on where you are right now
▸ If you walk but do no strength training — that is exactly the spot where this data moved the most. The amount you start with can be smaller than you think.
▸ If you have not started because there is no time — a difference showed up even among people doing about 12 minutes a week. This is not a story that begins only after three gym sessions a week.
▸ If you are already doing more than two hours a week — this data gives you no reason to expect that adding more improves this particular measure. Strength and physical function are a separate axis, and that is not what these studies measured.
▸ If you do neither aerobic nor strength work — in this data the aerobic side moved the numbers far more. Walking first is the right order.
The question gets easier if you change it. Not "what is the optimal number of minutes" but "am I above zero right now." The two datasets answer the first question differently. They answer the second one the same way.
How many steps you need is something I went through separately.
→
barosit.com/community/p/d5f1a8c3-9e46-4b28-8c73-1a6f3d9e5b20
Breaking it into short pieces came up in the post about stairs.
→
barosit.com/community/p/c8e4a1d3-7b56-4f29-a640-2d9b7e5c1f38
One thing nagged at me the whole time I was writing this. What these studies measured is exercise done separately in your free time, which is a different axis from the hours you spend sitting.
What BaroSit does is tell you when you have been locked in one posture too long, so it is no help at all in getting you started on strength training. To borrow this post''s own conclusion, those two do not substitute for each other either.
→ Have a look at
barosit.com if you are curious.
The evidence on sitting time and movement is collected at
barosit.com/science.
Sources
• Momma, Kawakami, Honda & Sawada, 2022 · Br J Sports Med 56(13):755–763 — Systematic review and meta-analysis of 16 prospective cohorts. All-cause mortality for any versus no muscle-strengthening activity, RR 0.85 (95% CI 0.79–0.93; 7 studies, 42,133 deaths, 263,058 participants; I² 83%). In the dose-response analysis (6 studies) the lowest RR of 0.83 (0.79–0.86) fell at 40 min/week, with estimates below 1.00 out to roughly 140 min/week. Cardiovascular disease 0.83 (0.73–0.93), total cancer 0.88 (0.80–0.97), diabetes 0.83 (0.77–0.89), lung cancer 0.90 (0.83–0.98); no association for colon, kidney, bladder or pancreatic cancer. The authors graded the all-cause mortality, cardiovascular and total cancer evidence as GRADE "very low," stated they could not test for publication bias because of the small number of studies, and noted the pooled estimates may be overestimated. The statement that the "at least twice a week" recommendation rests primarily on musculoskeletal evidence is also from this paper. No competing interests declared
• Zhang, Lee, Rezende, Ma & Giovannucci, 2026 · Br J Sports Med 60(12):874–883 — Three US cohorts (Health Professionals Follow-up Study, Nurses'' Health Study, Nurses'' Health Study II), 147,374 participants, up to 30 years of follow-up, 35,798 deaths, median baseline age 54. Strength training assessed biennially and modelled as a cumulative average. Adjusted for aerobic activity, hazard ratios for all-cause mortality were 0.95 (0.92–0.97) for 1–29 min/week, 0.91 (0.87–0.95) for 30–59, 0.91 (0.86–0.96) for 60–89, 0.87 (0.81–0.95) for 90–119 and 0.92 (0.86–0.98) for 120 or more. The per-band averages quoted above (12, 42, 72, 102 and 204 min) are the age-adjusted means in Table 1 of that paper. In the joint analysis, people below 7.5 MET-hours/week of aerobic activity stayed at 0.89–0.97 when strength training was added, those in the 15–45 MET-hour range fell as far as 0.53–0.62, and those at 45 MET-hours/week or above sat at 0.53–0.58 regardless of strength training. The under-55 subgroup rested on 624 deaths: 0.73 (0.56–0.94) for 1–29 min/week and 0.85 (0.68–1.05) for 30 or more. The authors note the questionnaire did not capture calisthenics or Pilates, and that 97% of participants were white. No competing interests declared
• Rahmati et al., 2025 · J Cachexia Sarcopenia Muscle 16(2):e13772 — Umbrella review of 48 meta-analyses. No association between physical activity or sedentary behaviour and mortality reached the top grade or the one below it. Resistance training on its own reached the third grade for cardiovascular mortality, and aerobic plus resistance training reached it for all-cause mortality. No conflicts of interest declared
Everything cited here is observational, and strength training time was self-reported by participants. Observational studies show associations; they do not establish cause and effect.
This post is general information, not medical advice. If you have heart disease, or have experienced chest pain, severe breathlessness or dizziness during exercise, please speak to a clinician before starting something new.